building an intent-based execution layer for AI workloads devs can run models without dealing with infrastructure.

github

productHunt

Benedict Gbogr  • Peerlist

<h3>AI workload execution platform.</h3>Jungle Grid is an AI workload execution platform that routes inference, training, and batch jobs across fragmented GPU capacity with explicit fit checks, health-aware placement, and recovery when nodes go bad. Stop juggling providers, GPU families, and fallback paths by hand.<ul><li>Submit jobs by intent: Describe the workload, model size, and optimization goal from the CLI, API, or MCP. Jungle Grid turns requests into placement decisions without making you guess GPU, storage, region, or provider combinations up front.</li><li>Reliable execution across fragmented GPU capacity: Placement decisions account for price, reliability, latency, queue depth, VRAM fit, and thermal state before dispatch.</li><li>VRAM Fit: Jobs that cannot fit current capacity are rejected explicitly instead of sitting pending forever.</li><li>Auto Requeue: When a node drops or goes stale, affected jobs are requeued onto healthy capacity automatically.</li><li>Compute network: Absorb fragmented capacity from managed providers and independently operated nodes into one execution surface.</li></ul>Key Features:<ul><li>Explicit fit checks before dispatch</li><li>CLI, API, MCP, and portal entrypoints</li><li>Automatic recovery from node failures</li><li>Routing across mixed GPU pools</li><li>Integration with MCP hosts</li></ul>